TWI465699B - Method of water level measurement - Google Patents

Description

Water level measurement method

The invention relates to a water level measuring method, in particular to a water level measuring method for calculating a water level height through image analysis processing technology.

Taiwan is located in the western Pacific monsoon region, which is extremely vulnerable to disasters. On average, it has been invaded by the typhoon in the western Pacific for more than four times a year. The heavy rains brought by the typhoon often lead to frequent floods and serious economic losses. In addition, with the global warming and climate change, the rainfall patterns have changed. In recent years, extreme rainstorms of heavy rainfall have been common in recent years. Extreme rainfall patterns are likely to cause urban areas to be unable to eliminate excess rainfall and areas in a short period of time. Flooding due to flooding is a serious threat to the lives and property of the people.

Accordingly, the monitoring and early warning of floods has always been the primary target of disaster prevention. The conventional flood monitoring method is usually after manual notification, and the relevant personnel are dispatched to the flooding site to measure the water level by a conventional contact water level measuring device. The measurement results are reported to the disaster response center for statistical treatment to determine the flooding situation, and then to issue a notification or early warning. However, this method often requires a lot of manpower to be used for the flooding site water level measurement and result aggregation statistics. Therefore, for low-lying areas with high flooding probability, the conventional contact water level measuring device is usually set up to monitor the water level at any time, and the measurement results are automatically returned by wired or wireless communication to save disaster relief resources.

However, since the conventional contact type water level measuring device has to be inserted into the water for measurement, it is often difficult to measure successfully when the water flow is in a hurry, and even there is a risk that the device is directly washed away or destroyed. In addition, the conventional contact water The measuring range of the position measuring device is often limited by the size of the device itself. When the flooding depth is too high, it is easy to have a situation in which the device itself is submerged and the measuring function is lost.

In order to solve the problems faced by the above-mentioned conventional flood monitoring method, for example, the content of the patent application of the "Lens Optical Image Water Level Measuring Device and Method" of the Republic of China No. 201024687 discloses a conventional non-contact water level measurement. The method can inject a water source into a body of water by a non-contact method, and take a water surface image to measure the water level of the water body. The device and method thereof of the patent can prevent the conventional contact type water level measuring device from extending into the water, and have the disadvantages of easy measurement failure or damage to the device, but the device still needs to be installed at each location to be detected, and still has a cost. The higher the problem, it is difficult to implement a wide range of flood monitoring. Moreover, since the water body tends to be turbid and full of debris when flooded, the method of irradiating with a laser light source also has the problem that the water level cannot be accurately measured.

In summary, the conventional non-contact water level measurement method can achieve the purpose of "measuring the water level in a non-contact manner", but there are still "the setting cost is too high" and "the measurement accuracy may be affected by the water color and Concerns such as the influence of drifting water in water, it is easy to derive different limitations and shortcomings in actual use, and there are inconveniences that need further improvement to improve its practicability.

The object of the present invention is to improve the above disadvantages, and to provide a water level measuring method, which only needs to analyze and process an image taken by a conventional surveillance camera, thereby measuring the water level height, and has the effect of reducing the system installation cost.

Another object of the present invention is to provide a water level measuring method, the water level The measurement method uses the image identification method to judge the height of the water level. The measurement is not affected by the color of the water body and the drifting material in the water, and has the effect of increasing the accuracy of the water level measurement.

Another object of the present invention is to provide a water level measurement method for detecting a boundary position by using a gradient operator edge search method, selecting a determination block in an image, and applying the gradient to the determination block. The operator edge search method has the effect of increasing the speed of the operation.

The water level measuring method of the present invention performs a water level measurement operation on an original image by a computer system, and the method includes: an image reading step, wherein the original image is read by the computer system, according to the size of the original image Calculating the actual size represented by a pixel; a grayscale processing step, performing a grayscale operation on the original image, the grayscale of the original image after the grayscale operation belongs to a grayscale colorscale range; an image enhancement step, adjusting The original image of the gray-scale operation is such that the gradation is evenly distributed between the maximum value and the minimum value of the gray-scale gradation range; an image correction step is performed according to the angle between the viewing angle of the monitoring lens and the horizon The image is rotated into a horizontal image, and a determination block is circled from the horizontal image as a post-production image; and a water level estimation step is performed by applying a gradient operator edge search method to the post-production image, and calculating the Image gradient and image boundary value of each pixel of the post-image, and a boundary projection value is estimated according to the image boundary value to estimate at least one boundary position; A position determination of the boundary surface location, and calculating the water level of the represented surface location based on the actual size of the pixel represented by the image of the original.

In the water level measuring method of the present invention, the method for calculating the image gradient and the image boundary value of each pixel of the post-image by the gradient operator sub-edge searching method is as follows:

Where G represents the image gradient, E represents the image boundary value, f(x, y) is the grayscale value of the pixel coordinate value of the judgment block (x, y), and Gx represents the horizontal direction of the judgment block. The gradient value, Gy represents the gradient value of the vertical direction of the judgment block, and ε is a threshold value.

The water level measuring method of the present invention, wherein the threshold value is obtained using a Canny edge detection operator.

The water level measuring method of the present invention, wherein the method of estimating the boundary projection value and estimating the boundary position is as follows: Where (x, y) represents the pixel coordinate value of the determination block, and I(y) represents a boundary projection value of a horizontal height in the determination block. When the boundary projection value exceeds a standard value, the horizontal height coordinate is set to a boundary position H.

The water level measuring method of the present invention, wherein the standard value is one-half of the width of the judgment block.

In the water level measuring method of the present invention, the method of determining the position of the water surface is to treat the at least one boundary position as the water surface position.

The water level measuring method of the present invention, wherein the square of the water surface position is determined The law takes the boundary position of the at least one boundary position that changes with time as the water surface position.

The water level measuring method of the present invention, wherein the water level measuring method further comprises a filtering processing step, after selecting a determining block in the image correcting step, filtering the determining block to eliminate noise, so that the The determination block forms the post-production image.

The water level measuring method of the present invention, wherein the filtering processing method used in the filtering processing step is intermediate value filtering.

The water level measuring method of the present invention, wherein the filtering processing method used in the filtering processing step is average value filtering.

In the water level measuring method of the present invention, the filtering processing method used in the filtering processing step is a Gaussian value wave.

In the water level measuring method of the present invention, the filtering processing method used in the filtering processing step is Laplacian value filtering.

The above and other objects, features and advantages of the present invention will become more <RTIgt; (pixels) is the smallest unit of an image used to represent the resolution of the image. For example, if the resolution of the image is 1024×768, it means that the image is common (1024× 768 = 786432) pixels are understood by those of ordinary skill in the art to which the present invention pertains.

The "color level" as used throughout the text of the present invention means the pixel The degree of shading of the displayed color component or brightness, for example, the range of the red (R), green (G), and blue (B) components of the color image is 0 to 255; or, gray The gradation range of the luminance of the gray-level image may range from 0 to 255, as will be understood by those of ordinary skill in the art to which the present invention pertains.

The "probability distribution" as used throughout the present invention refers to the probability distribution of the color gradation range of all pixels in an image, that is, the distribution of the gradation of each pixel of the image, for example, the color of a grayscale image. The order range is 0~255 (that is, 256 levels), then the probability distribution includes the following information: the number of pixels with gradation of 0, the number of pixels with gradation of 1, the number of pixels with gradation of 2, and the color gradation A number of pixels of 255 is understood by those of ordinary skill in the art to which the present invention pertains.

Please refer to FIG. 1 , which is a system architecture diagram of a preferred embodiment of the water level measurement method of the present invention. Wherein, a computer system 2 (for example, a conventional computer host, a file server, or a cloud server) is connected as an execution architecture by at least one monitoring lens 1 (for example, a conventional surveillance camera, a network camera, or a nighttime infrared camera). The at least one monitoring lens 1 can capture an original image A, such as a single or continuous image, and the original image A can be a color or grayscale image. The original image A There are several pixels, each pixel has a color gradation, and the range of values that can be represented by the color gradation is the gradation range of the image. The computer system 2 is coupled to the at least one monitoring lens 1 to receive the original image A, and according to the operation flow disclosed in the preferred embodiment of the water level measuring method of the present invention, the at least one monitoring lens 1 can be measured. The height of the flooded water level in the area. In this embodiment, the original image A is made in a single color image. For the following description of the implementation, but not limited thereto, and the like, it can be applied to the water level measurement of black and white or continuous image, which can be understood by those having ordinary knowledge in the technical field to which the present invention pertains. Narration.

In addition, an image server 3 is preferably connected in series between the at least one monitoring lens 1 and the computer system 2 for receiving and collecting the original image A captured by the at least one monitoring lens 1 and performing time sampling processing or imaging. The compressed image is processed, and the processed original image A is transferred to the computer system 2. Thereby, it can be avoided that when the number of the at least one monitoring lens 1 is too high, or the length of the original image A taken is too long, the formed image file size is too large, causing the computer system 2 to be unloaded.

Please refer to FIG. 2, which is a flow chart of the operation of the preferred embodiment of the water level measuring method of the present invention. The water level measurement method includes an image reading step S1, a grayscale processing step S2, an image enhancement step S3, an image correction step S4, a filtering processing step S5, and a water level estimation step S6, respectively .

In the image reading step S1, the original image A is read by the computer system 2, and in the embodiment, the original image A is an RGB image, and the gradation range of the original image A is 0 to 255. However, the invention is not limited thereto. The original image A is recorded by a monitoring lens 1 and the monitoring lens 1 is disposed at a fixed position. Therefore, the actual size represented by the original image A can be preset in the computer system 2. The image reading step S1 then calculates an image scale ΔH according to the pixel size of the original image A and the actual size represented by the original image A, which can be used to convert the actual size represented by the arbitrary pixel value. For example, suppose that the original image A taken by a surveillance lens 1 includes a commander station, and the actual height of the commander is known to be 500 cm. If the computer system 2 reads the original image A and measures the height of the command platform to be 100 pixels, it can be inferred that one pixel represents 5 cm, and the image scale ΔH is 5 (cm/pixel). When the image reading step S1 is completed, the grayscale processing step S2 is started.

The grayscale processing step S2 performs gray scale processing on the original image A. The main principle is to convert the original image A tones to the color according to the color gradations of the red, green and blue components of each pixel of the original image A. The tone scale of the grayscale image is in the range of 0 to 255. The tone conversion mode is preferably as shown in the following formula (1): f(x, y) = 0.299 × R (x, y) + 0.587 × G (x, y)+0.114×B(x,y) (1) where (x,y) is the pixel coordinate value, f(x,y) is the color scale of the grayscale image, R(x,y), G(x, y), B(x, y) represent three color gamuts of red, green and blue. The original image A is converted to form a grayscale image, and the color is relatively easy to process.

The image enhancement step S3 performs image enhancement processing on the grayscale image to obtain a enhanced image. The main principle is that the gray scale image probability distribution of the grayscale image is averaged to 0~255, thereby strengthening the grayscale image. Contrast value. For example, if the probability distribution area of a grayscale image is between 25 and 155, which is a case where the screen is dark, the probability distribution area can be adjusted to 0 to 255 via the image enhancement step S3, so as to facilitate subsequent steps. The details are interpreted. Since the picture quality of the original image A captured by the monitoring lens 1 is very susceptible to weather, time, and the like, resulting in image blur, noise generation, and overall hue is too bright or too dark, the image is enhanced through the image enhancement step S3. After the intensive treatment, it is more accurate to observe the changes in the details.

Referring to FIG. 2 and FIG. 3, FIG. 3 is an internal flowchart of the image correcting step S4 of the preferred embodiment of the water level measuring method of the present invention. The image correcting step S4 includes a horizontal correcting sub-step S41 and a The judgment block circle selection sub-step S42. The horizontal correction sub-step S41 is to perform horizontal correction on the enhanced image to obtain a horizontal image. Since the angle of view of the at least one monitoring lens 1 is not directly horizontally looking at the object to be detected, it is necessary to correct the level of the object to be detected. To facilitate the detection of flooding depth. Referring to FIG. 4 again, the at least one monitoring lens 1 is disposed at a fixed position. Therefore, the computer system 2 has preset a horizontal correction angle θ for any of the monitoring lenses 1, and the horizontal correction angle θ The angle between the viewing angle V of the monitoring lens 1 and a horizon G is used to correct the image from each of the monitoring lenses 1 by using the horizontal correction angle θ, and the horizontal correction mode is as shown in the following formula (2): x ' = x cos(θ)+ y sin(θ) y' =- x sin(θ)+ y cos(θ) (2) where (x,y) is the pixel coordinate value of the enhanced image, (x', y') is the pixel coordinate value of the horizontal image. The horizontal correction sub-step S41 is selectively executable. For example, when a monitoring lens 1 is directly viewed from the object to be detected, the enhanced image is a horizontal image. The computer system 2 presets the horizontal correction angle θ to be equal to zero. That is, the horizontal correction sub-step S41 is omitted. The determining block circle selection sub-step S42 selects a determination block R from the horizontal image, and the size of the determination block R is M×N pixels, that is, the horizontal width of the determination block R has a total of M pixels. The length of the vertical direction has a total of N pixels, and the size of the determination block R preferably accounts for more than five percent of the original image A. In addition, the determination block image R includes a horizon G, wherein the actual height of the horizon G is H _t (cm), and the horizontal height coordinate value of the horizon G is H ₀ , that is, the horizon G is located. The judgment is the H _0th pixel of the vertical position in the block R, and 0 ≦ H ₀ <N.

The filtering processing step S5 performs filtering processing on the R portion of the determining block, and the main principle is to perform a detection comparison on the determining block R by using a mask, which is a conventional median filter ( Median filter), and the size of the mask is preferably 1×(M/2), and M/2 is half of the width of the determination block R, and the pixel of the determination block R is covered by the mask. Replaced by the intermediate value. Thereby, the noise of the determination block R can be reduced, and since the water level usually only rises or falls horizontally, the filtering process does not filter out the water level information component, and the determination block R can be further provided through the filtering process. Evaluate the water level efficiently. Through the above steps, the continuous image processing process can be performed on the determination block R of the original image A. The filtering processing step S5 finally generates a post-production image B, which is a size of M×N pixels and color. A grayscale image with a range of 0 to 255 for subsequent water level estimation. The filtering processing step S5 is selectively executable. For example, when an original image A is sufficiently clear and does not contain noise that may affect the operation result of the subsequent step, it is not necessary to filter the determination block R selected in the original image A. Processing, so this step can be omitted, and the determination block R is a post-production image B.

Referring to Figures 2 and 5, FIG. 5 is an internal flowchart of the water level estimating step S6 of the preferred embodiment of the water level measuring method of the present invention. The water level estimating step S6 includes a boundary estimating sub-step S61 and a boundary. a projection sub-step S62, a water level determining sub-step S63 and a result calculating sub-step S64, the boundary estimating sub-step S61 is a gradient operation sub-edge searching method, the main principle is to calculate the image gradient of each pixel in the post-image B (gradient) the value G, and calculate the image boundary value E of each pixel according to the image gradient value G. The image boundary value E is 1 or 0, which represents that a pixel is a boundary or not a boundary. The calculation method of the image gradient value G and the image boundary value E is as shown in the following equations (3), (4) and (5):

Where f(x, y) is the gray scale value when the pixel coordinate value of the post-image B is (x, y), as shown in the above formula (1), and 0≦x<M and 0≦y <N; Gx represents the gradient value of the horizontal direction of the post-image B, Gy represents the gradient value of the vertical direction of the post-image B, and ε is a threshold. When the image gradient value G of one pixel (x, y) exceeds the threshold ε, the image boundary value E(x, y) of the pixel (x, y) is set to 1; otherwise, the pixel is set (x, y) The image boundary value E(x, y) is 0. The threshold is preferably determined using a Canny edge detector.

The boundary projection sub-step S62 is to accumulate a horizontal boundary image boundary value E in the post-image B to generate a boundary projection value I, and determine whether the horizontal height is a boundary position according to the boundary projection value I. . The projection accumulation method and the boundary position determination method are as shown in the following equations (6) and (7):

Wherein, since (x, y) is the pixel coordinate value of the post-production image B, y is a horizontal height coordinate of the post-production image B, E ( x , y ) is the sum of all image boundary values E of the horizontal height coordinate y, that is, the boundary projection value I(y) of the post-image B at the horizontal coordinate y. When the boundary projection value I(y) exceeds a standard value, the horizontal height coordinate y is set to a boundary position H. In the embodiment, the standard value is preferably the width of the post-production image B. One of the points (M/2).

If the post-image B generates only a boundary position H in the boundary projection sub-step S62, the water level determining sub-step S63 directly sets the boundary position H to a water surface position. However, in addition to the water surface, the post-production image B may have other objects determined as boundary positions H, such as passers-by, vehicles or other obscured objects, etc., resulting in the boundary position H generated by the boundary projection sub-step S62. The number is more than one. Therefore, in order to estimate the water level more accurately, the water level determining sub-step S63 can analyze and compare the plurality of boundary positions H. The main principle is to compare the original image A received by the computer system 2 at different times, and the different time is received. After the original image A is generated, the image B is generated by the boundary projection sub-step S62, and a plurality of boundary positions H are respectively generated. Since the water level generally rises or falls with time during flooding, the water surface should be in each of the post-images B. By changing the boundary position H, by comparing the plurality of boundary positions H, the boundary position H maintained at a fixed height can be excluded, and only the boundary position H that changes with time is saved. And set to a water surface position.

The result calculation sub-step S64 compares the water surface position with the horizontal height coordinate value H ₀ of the horizon G, and according to the actual height at which the horizon G is located, Ht (cm) and the image scale ΔH (cm/ Pixel), calculate the actual height represented by the water surface position. For example, when the foregoing step determines that a boundary position H is a water surface position, the actual height represented by the water surface position is H _t +(HH ₀ )×ΔH(cm), which is taken by the monitoring lens 1. The height of the flooded water level in the area.

In summary, the water level measurement method of the present invention starts from the image reading step S1, in which the computer system 2 reads an original image A, through the grayscale processing step S2, the image enhancement step S3, and the image correction step. Subsequent steps such as S4 and filtering processing S5, performing image analysis processing on the original image A to generate a post-production image B, and then performing the water level estimating step S6 to obtain a water level height, which is the flooding in the original image A. height. Accordingly, the invented water level measurement method can achieve the purpose of calculating the water level height through image analysis processing technology.

Therefore, the preferred embodiment of the water level measuring method of the present invention only needs to measure the water level height by using the original image A provided by the monitoring lens 1 in conjunction with a computer system 2, and the cost is compared with the conventional contact type or non- The contact type water level measuring device is greatly reduced, and the monitoring lens 1 is a conventional monitoring camera or a network camera, and thus can be installed in a large amount for large-scale monitoring.

Furthermore, the preferred embodiment of the water level measuring method of the present invention uses an image recognition method to identify the relative relationship between the water level and the surrounding scenery in an image, thereby finding the water surface position and calculating the water level height, thereby accurately measuring the flooding. The water level at the time is not affected by the color of the water and the drifting water.

In addition, in the preferred embodiment of the water level measurement method of the present invention, a determination block R is taken out from the original image A, and a post-production image B is generated according to the determination block R, and a gradient operation is performed on the post-production image B. The sub-edge search method is used to estimate the boundary detection. Compared with the unselected decision block R, the gradient operation sub-edge search method is directly applied, which saves a large number of calculation processes. Therefore, the operation speed can be greatly improved.

In the preferred embodiment of the water level measuring method of the present invention, the water level in the image can be quickly and effectively measured by analyzing and processing an image, thereby improving image restoration quality and compression efficiency, thereby achieving "lowering" System setup cost, "increasing water level measurement accuracy" and "improving operation speed".

While the invention has been described in connection with the preferred embodiments described above, it is not intended to limit the scope of the invention. The technical scope of the invention is protected, and therefore the scope of the invention is defined by the scope of the appended claims.

〔this invention〕

1‧‧‧Monitor lens

2‧‧‧ computer system

3‧‧‧Image Server

A‧‧‧ original image

R‧‧‧Judgement block

B‧‧‧post image

S1‧‧‧Image reading step

S2‧‧‧ Grayscale processing steps

S3‧‧‧ Image Enhancement Steps

S4‧‧‧Image correction procedure

S41‧‧‧Level Correction Substep

S42‧‧‧Judgement block circle selection substep

S5‧‧‧Filter processing steps

S6‧‧‧ water level estimation steps

S61‧‧‧Boundary estimation substep

S62‧‧‧Boundary projection substep

S63‧‧‧ Water level determination sub-step

S64‧‧‧Result calculation substeps

V‧‧‧Viewing the angle of the lens

G‧‧‧Horizon

Θ‧‧‧ horizontal correction angle

H ₀ ‧‧‧Horizontal height coordinate value

H _t ‧‧‧The actual height of the horizon

H‧‧‧Boundary position

△H‧‧‧image scale

Figure 1 is a system architecture diagram of a preferred embodiment of the water level measurement method of the present invention

Figure 2 is a flow chart showing the operation of the preferred embodiment of the water level measuring method of the present invention

Figure 3: Internal flow chart of the image correction step of the preferred embodiment of the water level measurement method of the present invention

4 is a schematic view of a monitoring lens of a preferred embodiment of the water level measuring method of the present invention

Figure 5: Internal flow chart of the water level estimation step of the preferred embodiment of the water level measurement method of the present invention

S1‧‧‧Image reading step

S2‧‧‧ Grayscale processing steps

S3‧‧‧ Image Enhancement Steps

S4‧‧‧Image correction procedure

S5‧‧‧Filter processing steps

S6‧‧‧ water level estimation steps

Claims (11)

A water level measuring method for taking an original image by a monitoring lens and measuring the water level of the original image by a computer system includes the following steps: an image reading step, the original reading is performed by the computer system Image, calculating an actual size represented by a pixel according to the size of the original image; a grayscale processing step, performing gray scale operation on the original image, and the color scale of the original image after being connected by gray scale belongs to a gray scale color An image enhancement step of adjusting the original image of the grayscale operation such that the gradation is evenly distributed between the maximum and minimum values of the grayscale gradation range; an image correction step if the monitoring lens is The angle between the angle of view and the horizon is not equal to zero, and the original image is horizontally rotated and corrected according to the angle; and a determination block is circled from the original image as a post-production image; and a water level estimation step is The post-production image is processed by a gradient operator edge search method, and the image gradient and image boundary value of each pixel of the post-image are calculated, and the boundary value is pushed according to the image boundary value. a boundary projection value for estimating at least one boundary position; determining a water surface position according to the boundary position, and determining the water surface position by taking a boundary position of the at least one boundary position with time as the water surface position, and The water level represented by the water surface position is calculated according to the actual size represented by the pixels of the original image. The water level measurement method according to claim 1, wherein the gradient operator edge search method calculates the image gradient and the image boundary value of each pixel of the post-image as follows: Where G represents the image gradient, E represents the image boundary value, f(x, y) is the grayscale value of the pixel coordinate value of the judgment block (x, y), and Gx represents the horizontal direction of the judgment block. The gradient value, Gy represents the gradient value of the vertical direction of the judgment block, and ε is a threshold value. The method for measuring a water level according to claim 2, wherein the threshold is obtained using a Canny edge detection operator. The method for measuring the water level according to item 2 of the patent application scope, wherein the method for estimating the boundary projection value and estimating the boundary position is as follows: Where (x, y) represents the pixel coordinate value of the determination block, and I(y) represents a boundary projection value of a horizontal height in the determination block. When the boundary projection value exceeds a standard value, the horizontal height coordinate is set to a boundary position H. The water level measuring method according to claim 4, wherein the standard value is one-half of the width of the determining block. The method for measuring a water level according to the first, second or fourth aspect of the invention, wherein the method for determining the position of the water surface is to treat the at least one boundary position as the water surface position. The method for measuring a water level according to the first, second or fourth aspect of the patent application, wherein the water level measuring method further comprises a filtering processing step, after selecting a determining block in the image correcting step, the determining area The block performs filtering processing to eliminate noise, so that the determination block forms the post-production image. The water level measuring method according to claim 7, wherein the filtering processing method used in the filtering processing step is intermediate value filtering. The water level measuring method according to claim 7, wherein the filtering processing method used in the filtering processing step is average value filtering. The water level measuring method according to claim 7, wherein the filtering processing method used in the filtering processing step is a Gaussian value wave. The water level measuring method according to claim 7, wherein the filtering processing method used in the filtering processing step is Laplacian value filtering.